September 2016
Volume 57, Issue 12
Open Access
ARVO Annual Meeting Abstract  |   September 2016
One year prediction of low vision in uveitis patients
Author Affiliations & Notes
  • Mia Klinten Grand
    Rotterdam Ophthalmic Institute, Rotterdam, Netherlands
    Department of Medical statistics and Bioinformatics, Leiden University Medical Center, Leiden, Netherlands
  • Hein Putter
    Department of Medical statistics and Bioinformatics, Leiden University Medical Center, Leiden, Netherlands
  • Tom Missotten
    Rotterdam Eye Hospital, Rotterdam, Netherlands
  • Koenraad Arndt Vermeer
    Rotterdam Ophthalmic Institute, Rotterdam, Netherlands
  • Footnotes
    Commercial Relationships   Mia Grand, None; Hein Putter, None; Tom Missotten, None; Koenraad Vermeer, None
  • Footnotes
    Support  The Netherlands Organisation for Health Research and Development : ZonMw Grant 842005005
Investigative Ophthalmology & Visual Science September 2016, Vol.57, 4123. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Mia Klinten Grand, Hein Putter, Tom Missotten, Koenraad Arndt Vermeer; One year prediction of low vision in uveitis patients. Invest. Ophthalmol. Vis. Sci. 2016;57(12):4123.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose : To developed a statistical model to predict chances of low vision in uveitis patients under treatment. Accurate assessment of the risk is highly relevant for these patients as uveitis is the leading cause of legal blindness in the working population in the western world.

Methods : The data consisted of both eyes from 240 uveitis patients who visited the Rotterdam Eye Hospital in the period from 1979 to 2015. Mean follow-up time and mean number of visits were 4.5 years and 20 visits, respectively. A logistic regression model was employed to assess the probability of low vision (<0.3 Snellen) one year after the first visit. In order to deal with separation problems of the logistic model and to account for the dependence between eyes we employed Firth correction and used the sandwich estimator for the variance. A list of the independent variables included in the model can be found in Figure 1. The same model was refitted for patients at different time points during follow-up to assess how the one-year prediction of low vision changes with time. The models’ predictive abilities were evaluated using the receiver operating characteristic (ROC) and area under the curve (AUC).

Results : Figure 1 shows that visual acuity at baseline is very predictive (log odds ratio=4.84) for low vision one year after the first visit. The effect size is even increasing over the first year of follow-up (Figure 2 A). The one-year predictive ability of the model at baseline is high (AUC=0.878) and further increases with follow-up time (Figure 2 B).

Conclusions : The models have a good predictive ability and a patient’s current visual acuity is an especially strong predictor for low vision.

This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.

 

Logarithm of the odds ratio (OR) of the independent variables and the intercept for the model at first visit to the hospital (follow-up time 0). LogMAR was transformed by the arctangent to reduce the effect of large logMAR values. The reference category is given in brackets, 95% confidence intervals are indicated by black lines and significant p-values (5%) by colour.

Logarithm of the odds ratio (OR) of the independent variables and the intercept for the model at first visit to the hospital (follow-up time 0). LogMAR was transformed by the arctangent to reduce the effect of large logMAR values. The reference category is given in brackets, 95% confidence intervals are indicated by black lines and significant p-values (5%) by colour.

 

Part A: Logarithm of the odds ratio of logMAR (transformed by the arctangent) and the Intercept as a function of follow-up time, with 95% confidence intervals. Part B: ROC curves and AUC for a subset of the models.

Part A: Logarithm of the odds ratio of logMAR (transformed by the arctangent) and the Intercept as a function of follow-up time, with 95% confidence intervals. Part B: ROC curves and AUC for a subset of the models.

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×